Zscaler
Also known as: zscaler.com, Zscaler Agentic SOC, Zscaler Agentic SecOps, Zscaler AI Protect, Zscaler AI Broker
Zscaler runs a zero trust security cloud that sits inline between users, workloads and apps, and sells two agent lines on top of it. Agentic SOC uses dozens of specialized AI agents to group alerts from Zscaler and third party tools into threat stories, triage them, recommend containment and run playbooks that block, isolate or restrict access, manually, automatically or after approval. AI Protect discovers, red teams and governs the customer's own AI agents, MCP servers and model traffic.
Zscaler sells the Zero Trust Exchange, a cloud security platform that inspects traffic inline for internet and SaaS access, private application access, branch and workload connectivity, data security and digital experience monitoring, and it secures more than 750 billion transactions a day. Zscaler, Inc. is based in San Jose and trades on Nasdaq, more than half of the Fortune 500 rely on it, and its customers include Siemens, United Airlines, T-Mobile and Deutsche Börse. The security operations line draws on Red Canary, the detection and response company Zscaler acquired in August 2025, which still sells as Red Canary, a Zscaler company.
Agentic SOC is the agent product for security teams. It pulls alerts from Zscaler services such as internet access, digital experience, data loss prevention, deception and MDR, and from third party endpoint, identity, email and cloud tools, into the Data Fabric for Security, then groups related alerts into prioritized threat stories with attack paths from initial access to impact.
Specialized agents write threat summaries, correlate signals beyond fixed rules, triage and validate indicators, map activity to MITRE ATT&CK and recommend containment matched to the incident's severity and business impact. Playbooks act through Zscaler's own inline controls, blocking or unblocking URLs, files and source IPs and isolating threats, and through third party SOAR workflows and two way ITSM ticketing, and each one can run manually, automatically or only after a person approves it. Every determination shows the evidence for and against it.
Zscaler MDR, Exposure Management, Deception and Threat Hunting sit in the same Agentic SecOps family. AI Protect covers the other side: AI Asset Management finds agents, models and MCP servers across clouds and more than 2,900 apps, Secure Access to AI applies intent based guardrails and data filtering to AI use, red teaming runs more than 5,000 attack scenarios against the customer's AI apps and MCP servers, and AI Broker sits inline on MCP and agent to agent traffic with an agent registry that enforces what each agent may reach.
Pricing is quote only, with platform bundles, add on modules and standalone products sold through sales. Zscaler holds SOC 2 Type II, ISO 27001 and ISO 27018 certifications, FedRAMP High for its internet and private access services, IL5 for private access and CMMC Level 2, runs separate GovClouds in the US, and controls admin access through role based administration and SAML for admins in ZIdentity. Customers have no choice of the models behind its agents.
It fits large enterprises and agencies already running Zscaler that want triage and containment tied to the controls that sit in the traffic path, and security teams that need to inventory and police the AI agents their own staff build. It fits less well for teams outside the Zscaler platform, buyers who want to choose the model behind each agent, or anyone who needs a list price.
Vendor details
Canonical URL
https://www.zscaler.com
Category
Security / SOC agent
Subcategory
Zero trust security cloud with an agentic SOC that triages and contains threats, and controls that discover, test and govern the customer's own AI agents
Funding status
Public company. Zscaler, Inc., headquartered in San Jose, California, trades on Nasdaq (ZS).
Company status
independent
Use cases & customers
Primary use cases
Target customers
Deployment options
Integrations
Agentic SOC ingests alerts from Zscaler Internet Access, Digital Experience, Data Loss Prevention, Deception and MDR plus third party endpoint, identity, email and cloud tools, launches third party SOAR workflows, forwards distilled incidents to a SIEM and keeps two way ITSM ticketing. Containment runs through Zscaler and third party inline controls. AI Protect watches MCP and agent to agent traffic, and OneAPI exposes the platform to automation with OAuth tokens and a Python SDK.
In practice
A bank's SOC receives an endpoint alert, an unusual sign in and a data loss alert for the same employee within an hour. Agentic SOC groups them into one threat story, triages it as credential theft, and runs an approved playbook that blocks the attacker's IP, restricts the user's access and opens a ServiceNow incident.
A retailer with a dozen internal AI agents uses AI Asset Management to find two MCP servers nobody registered, runs red team scenarios against its customer service agent, and puts AI Broker in front of MCP traffic so the agent can no longer reach the payroll system.
A federal agency already on Zscaler's FedRAMP High internet access service lets low severity phishing playbooks run automatically while keeping high impact isolation steps behind analyst approval.
Sources & related URLs
Related / legacy domains
Research sources
Agentic Index coverage score
10.0 / 14 capabilities · 71%
| Integrations & Tool Calling | Full |
|---|---|
|
Agentic SOC takes in alerts from endpoint, identity, email and cloud tools through prebuilt connectors and acts back out through them: it launches workflows in a third party SOAR, opens and updates tickets in the customer's ITSM in both directions, and forwards distilled incidents to the SIEM. Containment runs through Zscaler's own inline controls and through third party inline controls, and the wider Agentic SecOps line connects to EDR, IAM, SIEM, SOAR and ITSM or CMDB systems. Sourcezscaler.com/products-and-solutions/agentic-socread 2026-11-18 |
|
| Workflow Orchestration | Full |
|
Dozens of specialized agents split the work on each incident: one set groups and correlates alerts, another triages indicators, another writes the threat summary and maps it to MITRE ATT&CK, and another proposes the response. Multistep playbooks then chain Zscaler controls and third party controls in one run, with AI suggesting the next best action and its likely business impact at each step, and guided step by step paths for analysts who lead the response themselves. Sourcezscaler.com/products-and-solutions/agentic-socread 2026-11-18 |
|
| Knowledge Grounding & RAG | Full |
|
The Data Fabric for Security is the standing layer the agents draw on: it unifies Zscaler and third party alerts, enriches them with business context, zero trust telemetry and posture details, and keeps the correlations that turn single alerts into threat stories. Exposure Management maintains the asset inventory and aggregated vulnerability findings, and ThreatLabz research and more than a decade of SOC and MDR casework shape how the agents read what they see. Sourcezscaler.com/products-and-solutions/security-operationsread 2026-11-18 |
|
| Human Oversight & Guardrails | Full |
|
Each playbook runs in one of three modes: manually under an analyst, automatically, or only after a person approves it, and teams are pointed toward switching on full automation as their confidence grows. Recommended containment is matched to the severity of the incident so that a low risk alert does not trigger a disruptive block. On the AI Protect side, AI Broker enforces what each of the customer's own agents may reach over MCP and agent to agent traffic, and intent based guardrails govern multi turn agent conversations. Sourcezscaler.com/products-and-solutions/agentic-socread 2026-11-18 |
|
| Security, Identity & Governance | Full |
|
Zscaler holds SOC 2 Type II, ISO 27001 and ISO 27018 certification, FedRAMP High for its internet and private access services, an IL5 provisional authorization for private access, CMMC Level 2 and FIPS 140-2 validated crypto modules. Admin access is role based: each admin account carries a role and a scope, organizations build predefined or custom roles with read or full rights per area, admins can sign in over SAML 2.0 through ZIdentity, and access to the admin portal can be restricted by source IP. Sourcezscaler.com/solutions/public-sectorread 2026-11-18 |
|
| Observability & Auditability | Full |
|
Every verdict the agents reach shows the evidence behind it, both what supports the call and what contradicts it. The investigation hub holds the logs, evidence and timeline for each incident, a visual alert explorer shows how alerts were grouped, and attack path maps lay the threat story out from initial access through lateral movement to impact, so an analyst can trace why the agents reached a verdict as well as what they concluded. Sourcezscaler.com/products-and-solutions/agentic-socread 2026-11-18 |
|
| Memory & State Persistence | Not documented |
|
Incidents persist as threat stories with their alerts, evidence and timelines in the investigation hub, and the Data Fabric keeps the correlated security data that later incidents are matched against. That is the security record itself rather than a memory the agents keep of their own, and analysts have no store of lessons, preferences or past decisions that the agents read back with its own scope and retention. Sourcezscaler.com/products-and-solutions/agentic-socread 2026-11-18 |
|
| Deployment & Data Residency | Partial |
|
The Zero Trust Exchange runs as a multitenant security cloud, and Zscaler also operates separate GovClouds in the US for federal work, approved for storing and processing federal tax information. Agentic SOC and AI Protect run inside that cloud, and neither comes with a region list, a dedicated deployment model or a residency option of its own for commercial customers. Sourcezscaler.com/solutions/public-sectorread 2026-11-18 |
|
| Prebuilt Agents / Templates / Packs | Partial |
|
The specialized agents for threat summaries, grouping and correlation, triage, recommended response and enrichment arrive built in and work as one product rather than as agents a team picks and deploys one by one. Containment playbooks are assessed for impact and recommended per incident, and AI Protect brings a library of more than 5,000 attack scenarios for red teaming across text, image, voice and documents. There is no catalog of agents a team can adopt for other jobs. Sourcezscaler.com/products-and-solutions/agentic-socread 2026-11-18 |
|
| Triggers & Channel Coverage | Full |
|
Alerts start the work. Signals from Zscaler Internet Access, Digital Experience, data loss prevention, Deception and MDR, and from third party endpoint, identity, email and cloud tools, stream into the Data Fabric for Security, where agents correlate and triage them as they arrive with no analyst kicking off the run. Exposure Management keeps an asset inventory current and reprioritizes findings from every source as new ones land, and Deception cuts off access automatically when a decoy is touched. Sourcezscaler.com/products-and-solutions/agentic-socread 2026-11-18 |
|
| Model Flexibility & Routing | Not documented |
|
The agents run on an agent first architecture trained on Zscaler's SOC, MDR and ThreatLabz experience. No model or provider behind them is named, and customers have no setting to choose one. AI Gateway and AI Broker route and govern the customer's own model traffic across providers, which is a control over the customer's AI use rather than a choice of model for Zscaler's agents. Sourcezscaler.com/products-and-solutions/agentic-socread 2026-11-18 |
|
| APIs / SDKs / MCP Extensibility | Full |
|
OneAPI opens the Zscaler estate to automation: teams create API roles and clients in the authentication service, obtain tokens with client secrets or signed JWTs, and call Zscaler APIs to build their own workflows, with a Python SDK, an authentication guide and rate limits on the Automation Hub. Its coverage centers on the zero trust services the agents act through, such as internet and private access policies, rather than on Agentic SOC's threat stories and playbooks. Sourceautomate.zscaler.comread 2026-11-18 |
|
| Testing, Debugging & Optimization | Full |
|
AI Protect red teams the customer's own AI applications and agents with more than 5,000 attack scenarios across text, image, voice and documents, mapped to MITRE ATLAS, NIST and the EU AI Act, then fixes the gaps it finds, and a dedicated mode tests MCP servers. A prompt hardening service sits beside it. Agentic SOC's own agents have no comparable test suite that customers run against them. Sourcezscaler.com/products-and-solutions/ai-securityread 2026-11-18 |
|
| Browser / Computer-use | Not documented |
|
The agents act through connectors and inline controls rather than by operating a browser or a desktop. Zero Trust Browser is a separate isolation product that keeps people's browsing away from their devices, and Cyber Isolation renders risky pages remotely; neither is a browser the agents drive to click through another company's interface on an analyst's behalf. Sourcezscaler.com/products-and-solutions/agentic-socread 2026-11-18 |
|
The Agentic Index coverage score grades every vendor Full, Partial or Not documented against the same 14 buyer facing capabilities, from public evidence only. Each capability links to how all vendors in the index score on it. How this evidence is graded
Pricing
Quote only; sold as platform bundles, add on modules and standalone products
Enterprise subscriptions built from platform bundles, add on modules and standalone products, with some modules sized by users, throughput, endpoints or systems
What is public
Zscaler sells by quote. The Essentials Platform and the Zscaler Platform are bundles of internet access, private access and standard security modules, with advanced modules and standalone products added on. Agentic SOC and AI Protect carry no list price.
Billing mechanics
Bundles cover core zero trust services, and add ons such as Deception Advanced, Unified Vulnerability Management Advanced and GenAI Security are sold in tiered editions. Some products, including Deception and Unified Vulnerability Management, can be bought without a platform bundle.
Cost watchouts
Agentic SecOps sits on Zscaler telemetry, so its value grows with how much of the Zero Trust Exchange a customer already licenses.
Variable cost rationale
Contracts are quoted subscriptions sized up front rather than metered by agent activity.
Additional watchouts
Agentic SOC and AI Protect do not appear as named line items in the plans.
Sales call required
Yes, required for paid access
Free / trial
No free plan or trial; demos on request
Lowest paid plan
None; every plan is quoted by sales
Commercial notes
Agentic SecOps runs without SIEM ingestion fees and forwards only high priority threats to a SIEM when needed.
Missing data
Zscaler gives no prices for any bundle, module or agent product.
Related vendors
- 7AI — Agentic SOC from the Cybereason founders: domain specialized agents…
- Abnormal AI — Behavioral AI email security with three named autonomous agents: a…
- AirMDR — AI native MDR whose AI analyst investigates every alert and writes…
- Airrived — Agentic OS that unifies SOC, GRC, IAM, vulnerability management, IT,…
- Andesite — Human-AI SOC platform where customers build their own agents and…
- Anvilogic — Agentic security operations platform that works on top of existing…
Alternatives to Zscaler
Other products that do the work Zscaler does, and what each one is.
- Capsule SecurityRuntime security layer for AI agents
- Arctic Wolf
- DrataTrust management platform
- Splunk
- ZeronCyber risk and security control plane
- AsteliaAI native exposure management platform